Veloria | Devpost

About Veloria Inspiration- Mental health support remains one of the most critical yet underserved needs in modern society. 65% of people struggling with mental health issues never seek help due to stigma, cost, accessibility barriers, or long wait times. We were inspired by the gap between the need for compassionate mental health care and its availability especially for students, working professionals, and those in underserved communities. We envisioned Veloria as a bridge: combining the empathy and evidence-based techniques of professional therapy with the accessibility and personalization of AI. The name "Veloria" evokes "valor" (courage to face mental health challenges) and "aura" (a calming, supportive presence). Our goal was to create a 24/7 AI therapist that doesn't just respond it understands, adapts, and empathizes.

What it does - Veloria is an AI-powered mental health companion that provides real-time, voice-based therapy sessions with a lifelike animated avatar. Here's what makes it unique: CBT-Inspired Therapy: Uses Google's Gemini 2.0 Flash to deliver Cognitive Behavioral Therapy techniques Emotion Detection: Analyzes your messages to detect anxiety, sadness, stress, or other emotions Context-Aware Responses: Remembers conversation history to provide personalized guidance Crisis-Aware: Recognizes when professional help is needed and provides appropriate resources Real-Time Voice Chat: Powered by ElevenLabs' Conversational AI WebSocket Empathetic Voice: Sarah speaks with a warm, compassionate voice that adapts to your emotional state Speech-to-Text: Automatic transcription of your spoken words Audio Level Visualization: Dynamic gradient background that responds to conversation intensity TalkingHead Technology: Realistic facial animations synchronized with speech Emotional Expression: Avatar expressions match the therapeutic context Immersive Experience: Creates a more personal, human-like therapy session Movement Suggestions: AI suggests micro-movements (breathing exercises, shoulder stretches) during sessions Biometric Processing: Future-ready to analyze heart rate and HRV for adaptive recommendations Stress Detection: Identifies elevated stress levels and adjusts therapeutic approach History Tracking: Review past therapy sessions and progress Session Summaries: Exportable session notes and insights User Authentication: Secure Google OAuth login to protect your data

How we built it- Veloria is built on a hybrid microservices architecture that combines the strengths of multiple technologies:

Next.js 15 with App Router for server-side rendering and optimal performance React 19 for reactive UI components TailwindCSS 4 + Framer Motion for stunning animations and responsive design Custom UI Components: Animated gradient backgrounds that react to audio levels Progressive blur effects for infinite logo sliders Feature cards with hover-triggered gradient effects Fading text streams for real-time transcription

Challenges we ran into - Synchronizing the AI's voice output with the avatar's lip movements while maintaining low latency. Making Gemini provide empathetic, CBT-appropriate responses without overstepping boundaries or giving harmful advice. Challenge: Managing multiple API keys (Gemini, ElevenLabs, NextAuth, Google OAuth) across development and production. Designing an interface that feels calming and professional, not clinical or intimidating. Running 4 separate services (Frontend, Python, Node, TalkingHead) and ensuring they communicate correctly. Running 4 separate services (Frontend, Python, Node, TalkingHead) and ensuring they communicate correctly.

Accomplishments that we're proud of - We successfully orchestrated 4 independent services working in perfect harmony something that typically takes large teams weeks to achieve. Our startup script makes running the entire stack. Unlike chatbots that feel robotic, Veloria's Sarah actually listens. The combination of Gemini's language understanding, ElevenLabs' natural voice, and our CBT prompt engineering creates conversations that feel genuinely therapeutic. The interface is production-ready with:

Zero dependency on external UI libraries beyond shadcn/ui Custom-built animated components (infinite sliders, progressive blur, gradient backgrounds) Accessibility features (keyboard navigation, ARIA labels) Responsive design from mobile to 4K displays The audio level visualization that adapts gradient backgrounds based on conversation intensity is technically sophisticated yet feels effortless. It required deep understanding of Web Audio API, FFT analysis, and animation performance optimization. We created 7+ documentation files (AUTH_SETUP.md, ELEVENLABS_WEBSOCKET_SETUP.md, QUICK_START.md, etc.) making Veloria actually usable by other developers—something many hackathon projects lack.

What we learned

Advanced WebSocket Programming: Learned ElevenLabs' Conversational AI protocol and how to manage bi-directional audio streaming Prompt Engineering: Discovered how to craft system prompts that balance empathy, safety, and therapeutic efficacy Audio Processing: Gained expertise in Web Audio API, FFT analysis, and real-time visualization Microservices Coordination: Mastered orchestrating Python, Node.js, and Next.js services with proper CORS, health checks, and error handling

What's next for Veloria

We’re just getting started with Veloria. Our goal is to make mental health support more immersive, accessible, and connected to real-world care.

Short-Term (Next 3–6 Months)

Launching the mobile app so students can get support anywhere, anytime. Integrating wearables like smartwatches to detect stress levels and suggest calming exercises. Improving emotion detection to recognize subtle tone and language changes for more human-like responses.

Mid-Term (6–12 Months)

Introducing an AR companion mode where Veloria can appear in your environment to guide mindfulness or study breaks. Creating VR wellness spaces that let users step into peaceful, immersive worlds for guided therapy sessions. Partnering with universities to give students free access and improve campus mental health resources. Expanding language support to make Veloria available globally.

Long-Term Vision

Partnering with governments, hospitals, and clinics to integrate Veloria into public and private healthcare systems. Working with therapists and researchers to validate Veloria as a certified digital mental health companion. Building AI crisis detection to connect users to professionals during emergencies. Developing a fully immersive AR/VR therapy hub for group sessions and guided mental wellness experiences. Our vision is simple — to create a world where anyone can say, “I need someone to talk to,” and Veloria is there to listen, understand, and care.